For my situation, the entire point of PCIe is to add Thunderbolt ports and fiber IO. Using a Thunderbolt port to go to PCIe will not get me more Thunderbolt bussing. Everything will be off the same thunderbolt buss from the computer.
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Mac Pro 2023 = Another Mac Pro 2013 crisis and failure.
- Thread starter sunny5
- Start date
- Sort by reaction score
For my situation, the entire point of PCIe is to add Thunderbolt ports and fiber IO. Using a Thunderbolt port to go to PCIe will not get me more Thunderbolt bussing. Everything will be off the same thunderbolt buss from the computer.
Ah, my mistake.
Is there any PCIe Thunderbolt cards out there? I believe the protocol is basically “PCIe in a cable” so that might be an easy way to expand there. I’m not an expert in this type of thing so I truly don’t know.
Thanks for the clarification
Not yet. But for the last MacPro, each GPU came with four Thunderbolt ports on two busses. Thus, my current machine has 12 ports of which 10 are used regularly.
2023 Mac Pro will sell as well as Apple expects it to be. They will get their forecasted margins.
Perfect is the enemy of good.
It will make the minority of earlier Mac Pro users unhappy.
Having read up on the PCIe versions I've had wanted Apple to use these versions and skip PCIe 4.0.
- 2019 PCIe 5.0 for 2020 5nm M1 chips
- 2022 PCIe 6.0 for 2023 3nm M3 chips
- 2025 PCIe 7.0 for 2025 2nm M4 chips
Doing so may give Apple a distinct raw performance advantage.
Perfect is the enemy of good.
It will make the minority of earlier Mac Pro users unhappy.
Having read up on the PCIe versions I've had wanted Apple to use these versions and skip PCIe 4.0.
- 2019 PCIe 5.0 for 2020 5nm M1 chips
- 2022 PCIe 6.0 for 2023 3nm M3 chips
- 2025 PCIe 7.0 for 2025 2nm M4 chips
Doing so may give Apple a distinct raw performance advantage.
Hmm that’s an interesting pickle. As opposed to the OP you seem to actually have real world critiques.
Do you mind filling in some details of your workflow to help us understand? I think unfortunately you may be in the bucket of workflows that Apple may longer be supporting.
There are no Pro Mac users. Anyone who is still on the Mac after well over a decade since the last remotely competitive workstation from Apple is here because they like MacOS and don't need truly high end performance. Apple hasn't even tried to offer an up to date, competitive tower since 2009/2010. However, in the meantime, even a Macbook Air can do things no 2010 and earlier workstation could do. And the truly pro world has moved almost completely to using render farms and cloud computing solutions, not high end workstations. Users left using the mac for work either do office type jobs that let them by their own hardware, audio, or some 2D art forms such as photo editing, some illustration type stuff, or small single editor video projects. Since even a macbook can do any/all photo and video work anyone would do on a mac, the MacPro exists mainly to cater to the remaining audio professionals. Those who have non-video PCIe cards, those for whom the M2Ultra is more than enough power for any imaginable workflow.
How does this seem like 2013? the real knock on the 2013 was the lack of expansion. This thing is expandable out the ass. SoCs are just better than discrete CPU, RAM and GPU. There is no way to overcome that. Given a much larger power and silicon budget you can brute force through the problem, but the SoC is becoming the new motherboard. To keep up with gains everything has to be direct silicon connected with super low latency and super high bandwidth. 2023 Mac Pro is leagues ahead of the 2019 in all CPU metrics, and can have more NVMe RAID storage, running theoretically faster than dual channel DDR4 in the 2019 Pro, using one or two PCIe slots (easily could have 32TB in RAID 0 with a theoretical 64GB/s throughput). The only place the 2023 falls behind is in GPU, but, again, no one doing actual 3D or heavy compute work still runs MacOS. Would it be nice if this machine tried to change that? sure, I guess. But did anyone honestly expect Apple to completely pull a 180 on a decade plus of highly successful erosion of the Mac's high end competitiveness?
The M2 Ultra is a lot more CPU than anything in the 2019 model. It does not keep up with Sapphire Rapids or the theortical Threadripper 7000 chips, which is lame. But those platforms hardly exist and the M2 Ultra will be competitive or ahead of the 7950x and 13900k.
It's $1000 more after 4 years, or inflation adjusted a flat price. For that, you get a much more powerful base configuration. 2019 Pro launched with an 8core Xeon, 32GB DDR4 RAM, and glorified RX580 GPU. 2023 launches with twice the RAM, with 20x the bandwidth. The new pro ships with 24 much faster cores, for likely 4x the CPU performance of the base 2019 pro. It ships with more TB ports, more open PCIe slots, the ability to drive more pixels and more displays, built in accelerators equivalent to 7 of the add on Afterburner card. To build a 2019 Pro up to the base specs of the 2023 Pro would cost $10s of thousands. Lack of extra GPU compute hurts, and is stupid. But in every other way, this MacPro demolishes the old one in value for money.
NVMe, SATA, Audio interfaces, Video capture cards, networking adpaters. Cards with future I/O standards (USB 5 or 6, next gen thunderbolt? etc.) There are tons of options. All my old PowerMacs and MacPros run out of PCIe slots by the time they are about 5 years old, so many new standards have come out and can be added for a couple hundred bucks or less rather than by buying a whole new tower.
It took you until now to see that Apple doesn't care about workstations? They made that clear around 2010. And have never pretended to do anything other than provide slightly soft cushions along the glide path ever since. Apple wants customers who buy high volume products like a new fully loaded MacBook Pro or iPhone regularly. It's not worth it to chase the tiny fraction of people who want higher end hardware but just won't leave the mac. Even if they priced the MacPro competitively, at Studio pricing or lower, and added back full GPU support for some reason, these would sell in the hundreds or thousands not in the millions of tens of millions. They charge a lot to recoup RnD costs and discourage people who want this type of hardware from sticking with the Mac.
More of a rant than providing a place for legit open-minded discussion, as the poll might indicate at first.
I use RAID 5 drives for storing video files. I backup each drive in duplicate with two other RAIDs. I have about 5 sets of these and to prevent accidental corruption, I don’t power drives that aren’t being used, which means that I can’t daisy chain drives outside the backups.
I then use SSD raid 0 drives for high speed editing. Those use another few ports.
Lastly, there are some USB devices that don’t work well on a hub. They need to go direct, such as when using a keyboard and need to hold down a keystroke during a restart.
In general, I need a lot of ports. I can prob make do with 8 but I’m hoping that Sonnettech or OWC will make a PCIe with extra thunderbolt ports and / or extra SSDs or HDDs.
A maxed out 2023 MP is around $10K which isn’t bad, provided it does the job.
It's fundamentally an M2 Ultra Mac Studio with PCI slots and a user-replaceable SSD, but at least it's good to know they are still presenting that option (compared to the 2013 Mac Pro offering no such expansion slots!)
Which games though? There are very few demanding games with high quality macOS ports out there. RE village would be interesting to look at, certainly Death Stranding when it arrives (even though it’s a slightly older game it’s technology is super impressive).
I am a 5,1 owner. Swapped out the CPU's, memory, upgraded to a 2.5" SSD then a pcie m.2 drive, upgraded to a 750 Ti then VII, did the Pixlas Mod and now run Monterey on OCLP throughout the years. What am I supposed to upgrade to? My 5,1 has stayed top of the line for over a decade. Will this 8,1 still be viable in 2033 or even 5 years down the line in 2028? Pcie expansion is great but when it's the only defining feature over the Studio that is $3000 USD cheaper what's the point? I'm better off going for a Threadripper Pro workstation or something, but I use both MacOS and Windows. There will never be another workstation like the 5,1 it seems, which is sad as I thought Apple was committed to the high end space after rising above the 2013 Trash Can debacle.
I think that market has shrank. I think the current crop of M1/M2 based laptops/desktops are so powerful now that the need for a dedicated Pro chip beyond the Ultra may not be justified any more. It's incredible what people are able to accomplish with the M1 Max/M2 Max in the MBPs.
I mean they sold, but they were miserable to use. My Dad had a 9980HK 15" MBP and literally kept a mini-freezer next to his desk full of ice packs he'd place under the laptop and swap every ~30 minutes so he could keep the thing functioning. And all he does is web development, he wasn't pushing it hard, just opening several browsers with multiple tabs for testing.
Also, pretty sure the M2 Ultra will not physically fit in the 16" MBP chassis without major modification. It's a huge chip and needs a ton of heatsink contact. There is a market for workstation replacement laptops, but Apple hasn't been in the big laptop game since the Mac Portable, and brining the ultra to a laptop would require a device in the 7+ lb, 1"+ thick range at the minimumWith the thermal improvements of the 2021/2023 MBP 16" enclosure it will do way way better in reducing any possible throttling. Although it will be hella loud.
The 99Wh battery would likely halve battery life with the M2 Ultra. 40Wh battery would make it barely 20% of its original battery life.
It wouldn't be accepted as workstation replacement though. Not enough PCIe slots.
I could see a 2023 MBP 16" M2 Ultra for $5.5k-6.5k.
Yep. There's preciois little market between the M2Ultra and the datacenter/renderfarm market. Could apple potentially sell a lot of SoCs to datacenter if they really pushed on that front? Sure. But they're the largest company of all time by focusing on consumer hardware, so there's little incentive to pivot to exclusively B2B type products
There are tons of PCIe thunderbolt cards out there. Some of them even work in the 5,1 Mac Pro, which is pretty neat! No idea if those cards also work in the 7,1 or 8,1 but I would bet they are supported by MacOS moreso than the 2009 era backplane of the classic Mac Pro, so they probably do work in modern mac Pros. Silly that Apple doesn't sell one themselves, another 6 or 8 TB ports by using one of the PCIe slots on the 8,1 would be worth a lot to some users for sure. Worst case, the Apple I/O cards will end up on secondhand market someday and I bet there's no reason you can't run two of them in an 8,1
I think this may be given another glance when Apple’s are so performant *at a fraction of the energy consumption* data centers (and Apple) may be interested in that space. That’s a little ways off, but I think Apple’s hellbent focus on reducing power consumption is going to add up eventually in a way that can’t be ignored.
Who knows though, they may develop exclusively for their own operated data centers as part of their push for neutral emissions 🤷♂️
If I could redo my Mac purchases I'd steer clear of them from 2015-2019.
Last Intel Mac be a 2014 model then skip to a M1 in 2020.
That era is exactly why Apple had to switch. It became clear to me that these chassis (ignore the keyboard, that’s all on Apple) were designed for where Intel promised they’d be on their roadmap in terms of performance at a certain TDP. Intel missed that promise repeatedly for the better part of a decade.
Do none of the PC PCie thunderbolt cards work in the 7,1 Mac Pro? Gigabyte Titan Ridge cards definitely worked in the 5,1 and gave 2 thunderbolt ports quite affordably.
Any of the Butterfly Keyboard Macs are not worth buying both for the keyboard & the 14nm Intel chip that got stuck on 14nm from 2014-2020.
If it wasn't for Intel's greed to cut cost then Apple would never had gone on to make their Mac chips.
It would have been nice if Apple transitioned to Apple Silicon a year (Nov 2019) earlier so those doing WFH would have greatly benefited moving from 14nm Intel chips to 5nm Mac chips between then and Mar 2020.
Instead people were forced to buy into the last Intel Macs.
I bought no new macs from 2012 to 2020. For a reason. By the time my 2012 rMBP was showing its age, it was clear the only reasonable replacement was a 2015 rMBP, the last before the ultra thin idiocy. And I drove that plus a 2009 MacPro upgraded to hell and back until the M series launched. The intel 16" MBP was not terrible though, I wouldn't have hated having a loaded 9980hk + 5600m model if the AS transition hadn't been so clearly on the horizon.
Yep - I wish I had skipped my 2019 MBP the butterfly keyboard is bad and the Touch Bar developed the dreaded flicker.
Looking at nVidia's skyrocketing valuation on the backs of ARM datacenter SoCs coupled with "AI" accelerator compute modules, it might be worth a look for apple, but I just don't expect Tim's Apple to branch that far away from consumer devices and services when they've made so much money in that niche.
I was aware of the AS transition when I bought my 2019 MBP 16" Core i7 but was assuming a modest ~20% performance bump.
It did not occur to me that the transition would be from 14nm to 5nm chips and that going with SoC would result improvements on latency, power consumption, battery life, performance per watt, raw performance and making Mac logicboards as tiny as iPhone ones were the net benefit.
If I knew that then I'd have ceased any purchase of Mac notebooks from 2015-2020.
Jumping from 12yo 32nm & 10yo 22nm Intel chips to 5nm M1 & M2 chips would have been glorious!
I'm waiting to replace my 2012 iMac 27" 22nm for hopefully a 2023 large iMac M2 Pro within 4 months.
Last edited: